Epicosm

Epicosm integrates Tweets from Twitter with epidemiological cohort datasets to enable digital phenotyping and longitudinal analysis of mental health and is implemented as a Python-based framework for automated collection and secure linkage of social media data.

Key Features:

• Data Collection: Automates regular gathering of Tweets from specified cohort participant accounts to produce high-resolution time course social media data.
• Secure Data Handling: Designed for deployment within a cohort data safe haven and enforces privacy and security practices to protect participant confidentiality.
• Integration with Cohort Data: Stores social media data in a structured database and links it to existing epidemiological datasets to support validation of digital phenotyping algorithms.
• Expandability and Robustness: Modular architecture supports customization and scaling and was co-designed with input from cohort leaders and participants.

Scientific Applications:

• Validation of Digital Phenotyping Algorithms: Provides linked social media and cohort ground truth data to assess and refine algorithms that infer psychological states or predict mental health outcomes.
• Longitudinal Studies: Supplies high-resolution temporal Tweet data for longitudinal analyses of mental health trajectories and their associations with social media activity.
• Cross-disciplinary Research: Enables integration of computational analyses of social media traces with traditional epidemiological and clinical data for interdisciplinary studies.

Methodology:

Regularly collect Tweets from a predefined list of cohort participants' accounts and securely store and link the resulting social media records to existing cohort datasets within a protected environment.